AI agents are already defying instructions and taking actions their creators never intended. Now, as the technology gains the ability to spend money, Mastercard is confronting a pressing question. How do you keep autonomous software under control?
The risks are increasingly tangible. In July, an experimental OpenAI agent escaped its testing environment and accessed systems belonging to AI platform Hugging Face. More recently, Anthropic disclosed that its Claude models had exploited website vulnerabilities and submitted a false homicide tip to Philadelphia police during testing.
For Mastercard president and chief technology officer Ed McLaughlin, these concerns are especially relevant as the payments giant prepares for a future in which AI agents can independently make purchases and transact with other machines.
“This probably is what I’m most excited about,” McLaughlin said in an interview with Fortune Next to Lead.
Consumers could instruct an AI assistant to monitor ticket prices and automatically purchase seats within a set budget. Mastercard’s Agent Pay technology is designed to facilitate those transactions, while Agent Pay for Machines allows autonomous systems to transact with one another within defined parameters.
But giving AI agents the authority to spend money also means giving them the power to make costly mistakes. An agent could book a nonrefundable flight for the wrong date while staying within its authorized budget.
That raises a thornier question for Mastercard and the broader financial industry. Who is responsible when an AI agent follows the rules but makes the wrong decision?
That challenge reflects a dilemma facing business leaders today as AI agents move from assisting employees and consumers to making decisions on their behalf.
How much independence should companies give these systems, and what safeguards need to be in place before they act?
McLaughlin says the key is knowing which AI agent is making a purchase, who permitted it, and what it’s allowed to buy. ‘”Part of it is bringing humans back into the loop,” he said. “So even if the agent goes rogue and tries to do something, you can say no.”
Whether those safeguards will persuade consumers to hand over their purchasing power remains to be seen. McLaughlin, however, expects agentic commerce to follow the trajectory of online shopping and mobile payments, which initially faced skepticism before becoming routine as consumers experienced their convenience firsthand.
As companies give AI agents more authority, executives face a new responsibility: leaders must decide what those systems can do, when humans should intervene, and who answers when something goes wrong.
McLaughlin says that at Mastercard, whose network spans 3.7 billion accounts worldwide, those controls are essential to moving quickly. His guiding philosophy is a pointed departure from Silicon Valley’s longstanding mantra: “Move fast and don’t break things.”
Ruth Umoh [email protected]
The post What happens if a rogue AI agent goes on a spending spree? Mastercard is preparing for that risk appeared first on Fortune.
